A Neural Network Model of Concept-Influenced Segmentation
Robert L. Goldstone · eScholarship (California Digital Library) · 2000
Several models of categorization assume that fixed perceptual representations are combined together to determine categorizations.This research explores the possibility that categorization experience alters, rather than simply uses, descriptions of objects.Based on results from human experiments, a model is presented in which a competitive learning network is first given categorization training, and then is given a subsequent segmentation task, using the same network weights.Category learning establishes detectors for stimulus parts that are diagnostic, and these detectors, once established, bias the interpretation of subsequent objects to be segmented. Concept Learning and PerceptionThe current research explores the influence that learning a new concept has on the segmentation of objects into component parts.Recently a number of researchers have argued that in many situations, concept learning influences the featural descriptions used to describe a set of objects.Rather than viewing perceptual descriptions as fixed by low-level sensory processes, this view maintains that perceptual descriptions are dependent on the higher-level processes that use the descriptions (Goldstone, Steyvers Spencer-Smith, & Kersten, 2000;Schyns, Goldstone, & Thibaut, 1998).Evidence for this view comes from the study of expert/novice differences (Lesgold et al., 1988), influences of acquired concepts on the interpretation of stimuli (Wisniewski & Medin, 1994), and influences of category learning on psychophysical measurements of perceptual sensitivity (Goldstone, 1994).